Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 26, 2026Berlin

How Price-Aware Are AI Agents? Rival Cuts and Legal Rules

Martin Kaiser · Team kaiser-data

Submitted to AI Collusion Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: How Price-Aware Are AI Agents? Rival Cuts and Legal Rules

Open in a new tab (opens in new tab)
Code (opens in new tab)
Share

AI agents increasingly set prices, often without anyone watching. We ask two questions. First, how price-aware are language models on their own: how do they react when a rival cuts its price and no rule constrains them? Second, how do they decide when legal rules limit when a price may change, such as platform offer holds or Germany's 2026 rule that fuel prices may rise only at noon, and can an audit still tell the model from the rule? We tested 16 open models from Chinese and US developers in the same simulated market against a scripted rival. Without rules, the models differ widely: some ignore a rival's price cut, others cut their own price by more than twice what economic theory predicts. The developer's country does not sort them, and smaller and larger versions of one family often behave differently. Under a three-period price hold, 6 of 10 models tested show a larger response to the cut, which is the pattern collusion audits search for. For one model we separate the two causes: the rule itself explains less than half of the effect, and the rest is the agent adapting its own proposals to the held price. Logging what an agent wanted to charge therefore does not clear it. For audits, we recommend recording proposed prices, published prices and rule state together, and testing the exact model under the rules it will face. Results concern audit validity in a synthetic market.

Reviews

Judging this Sprint?

Review “How Price-Aware Are AI Agents? Rival Cuts and Legal Rules”

Your public critique appears on the project page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on the project page.

Shown on the project page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; the project page shows only that a judge declined, not why.

How much would this matter for the field if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.
Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

See our Privacy Policy.

  1. This is a valuable and insightful project, particularly given nascent regulatory proposals that AI pricing be made auditable by design. The insight that punishment by an agent may be a product of execution rule rather than collusion is interesting and potentially novel, and suggests investigation of execution rules.

  2. This project studies the effect of externally imposed rules that restrict the timing of price updates on LLM pricing agents. The rules are modeled after real-world cases. The study shows that such rules can cause widely different behavior among the LLM agents. One easy improvement this project can make is to test more sets of market parameters. The report for the project is clear overall. However, it sometimes uses inconsistent terms. For example, the Hold rule and Noon rule are sometimes referred to as Rule 1 and Rule 2, which makes it hard for the readers to catch.

  3. Nice diagnostic piece of work here. Showing how a regulatory rule (like a mandatory price-hold window) can mechanically make a non-collusive agent look like it's running retaliatory price cuts is a good cautionary tale for people designing empirical screens. The open-loop replay trick you used to separate the market rule from the agent's policy is clean and something other groups should borrow.

Cite this project

@misc{kaiser2026priceaware,
  title = {{How Price-Aware Are AI Agents? Rival Cuts and Legal Rules}},
  author = {Martin Kaiser},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/how-priceaware-are-ai-agents-rival-cuts-and-legal-rules-0w8g}},
  url = {https://apartresearch.com/sprints/projects/how-priceaware-are-ai-agents-rival-cuts-and-legal-rules-0w8g}
}

More from AI Collusion Research Sprint

All 392 entriesBrowse all projects

A Sprint is running now

AI x Epistemics Research Sprint · Nov 13 - 15, 2026